From my point of view, the paper is well written and the data, methodology and results are scientifically sound. Even if there is not so much technological novelty involved in the step from HOLIMO to SmHOLIMO, the technical improvements are very helpful for the cloud and aerosol physics community. One could argue that an inline holographic imager stays an inline holographic imager if just the optical properties are modified. The key point from my perspective is that the smaller pixel size enables the operator to probe a more substantial part of the left tail of the cloud droplet size distribution, which was one of the major weaknesses of holographic imagers in the past. This is extremely important for obtaining bulk cloud microphysical properties and performing intercomparison studies between passive and active remote sensing instruments and in-situ measurements.

However, I would like to have a bit more discussion about the sizing accuracy in the context of Figure 5, in particular for the panels q and u. For me it looks like the HOLIMO captured some of the small particles, but the sizing is incorrect (due to resolution limit effects). 

Besides that, I have only a few minor comments and recommend publication of this manuscript after consideration of my comments. 

Line 37: There is more literature available about the application of in-situ holography via tethered balloon systems, for example the setup by MPI-DS Goettingen on research vessel Maria S. Merian. Please add reference Stevens et al. (2021). 

Section 3.2.: I have some questions regarding the calibration. Could you please add information about temperature and relative humidity and also discuss the particle travel time between the outlet of the VOAG and the sample volume of both SmHOLIMO and APS? This is important to understand if evaporation could have changed the droplet size and might have introduced a bias in the calibration process. 

References:

B. Stevens et al.: EUREC4A, Earth System Science Data, 13, 4067–4119, https://doi.org/10.5194/essd-13-4067-2021, 2021.

General comments

This is not a novel instrument design. The authors describe a simplified digital, in-line, holographic instrument that operates in the design space where the ”typical resolution” is improved from 6µm to 4µm by reducing the sample volume by roughly a factor of 30.  The sample volume of SmHOLIMO is 0.5 cm³; whereas, the two other operational digital inline holographic instruments for cloud study, HOLODEC-II and HOLIMO3B, have sample volumes of 15 cm³.  The SmHOLIMO instrument’s ability to resolve “seemingly marginally smaller” particle diameters comes at the cost of limiting the instrument's ability to address other important cloud studies.  So the key question becomes, do the authors demonstrate that design trade was worth it?  

To help answer this, it would be helpful to have further discussion about the instrument design trades.  Something akin to Spuler and Fugal 2011, describing the HOLODEC instrument that was used in Beals et al., 2015; Larsen et al., 2018; Glienke et al., 2020.  In that case, it was clear that design choices were made to enable study of fine scale cloud droplet clustering and non-uniformity down to cm scales.  In this manuscript, it seems that this key advantage of holography was given up to gain a slight advantage in resolution.  Does it enable some new novel science?  For the limited case shown, it appears advantageous, but somewhat unclear why this instrument would be used versus others.

Specific questions

Line 11:  SmHOLIMO has the “potential for spatial analyses”?  With such a small volume what are the scientific problems that could be addressed?

Line 18: “We unequivocally show the importance of capturing the lower tail of the CDSD, which significantly affects the derived quantities”   In section 4.3, the authors show that the 2.3 µm improvement in resolution provides a more complete picture of the Cloud Droplet Size Distribution.  But for the derived projects, what happens when including the camera pixel uncertainty of 1.4µm (a large uncertainty when talking about an improvement in resolution of 2.3µm)  In other words how much certainty do you have in these derived products: Cloud Droplet Number Concentration (CDNC), Liquid water content (LWC), and Cloud Optical Depth?  Especially since in Line 5 they state:  “precise and accurate in situ measurements of cloud doppler size distribution, especially of cloud droplets smaller than 6 μm, are still lacking. This can lead to uncertainties in the microphysics and thus in weather and climate models, which are based on parameterizations often derived from these measurements”

Line 33-34:  After introducing in-line holography, the authors list references using photographic plate holography (e.g., Kozikowska et al. 1984, Borrmann and Jaenicke 1993)  They appear to be claiming that photographic plates are “a promising in situ approach” with a “robust setup”?  It would be helpful to make a distinction between those early precursors and operational digital holographic methods. 

Line 53: The authors state “The main reason that the resolution of holographic imagers is limited to around 6 to 10 μm is that they are designed to reliably observe both liquid and ice phase particles. Since under natural conditions ice crystal concentrations are usually several orders of magnitudes lower than cloud droplet concentrations (a few per liter vs. a few hundred per cm3).”  HOLODEC and HOLIMO instruments have a 6µm resolution imposed by a pixel resolution limit (Spuler and Fugal 2011, Ramelli et al 2020).  These instruments could resolve smaller particles by simply switching to higher resolution sensors (i.e., cameras with smaller pixels within the same frame size).   So it appears to be a design trade choice.  The reason for NOT pushing to higher resolution while maintaining large volumes exposes the much more serious limitation with digital holography – the extremely large computational expense to process the data. 

Line 67: The authors state that the instrument was designed to maintain “.., a large enough sample volume for sufficient counting statistic [sic] for large droplets.”  Yet, for the work here, it required averaging 5 volumes (1 second integration at 5Hz)  as discussed on Line 194 “To achieve a more accurate counting statistics the data was averaged to 1 s”.  So what volume is needed to have sufficient counting statistics?  In the quest for higher resolution, this instrument appears to have given up the ability to enable measurements within a single volume (for example, understanding the interaction of droplets on relevant length scales).

Line 111: “..the data transfer speed of the single-board computer limits the frame rate in field operations to 10 fps”  , yet on Line 192 it says “The SmHOLIMO data was analyzed with a frequency of 5 Hz (5 holograms per second)”  Why the change?

Line 128: “A uniform optical resolution across the whole sample volume is important to avoid a sampling bias linked to the size of cloud droplets”.  Please explain this further. One can divide up the sample volume however desired with holography.  Seems you could correct for this since you have measured the resolution limit of the instrument.

Line 146: “To obtain accurate cloud droplet diameters a calibration of the particle sizing algorithm is required.”  Holography does not require calibration.  In fact, this exact “calibration” process was done in Ramelli et al. 2020 and they state “ no correction to the sizing algorithm was made, because all size measurements agree within the square root of the pixel size (√3.01 μm = 1.73 μm).”   Nearly the same statement is repeated in this manuscript on line 167.  So this is not a calibration – albeit perhaps useful as  “sanity check” and a way to fine tune a brightness threshold.

Line 237: “There are two interrelated reasons for this underestimation”   The authors are suggesting a possible explanation for the discrepancy.  This was not experimentally verified to be the actual cause. 

General comments:

The paper “Putting the spotlight on small cloud droplets with SmHOLIMO – A new holographic imager for in situ measurements of clouds” by Christopher Fuchs and colleagues demonstrates the capabilities of the smHOLIMO instrument. While the instrument design itself is not novel—it follows the same layout as other in-line holographic probes—a better camera and smart sample volume selection enable it to measure droplets as small as 3.7 microns, compared to the previous 6-micron limit.  This improvement makes SmHOLIMO comparable to existing forward scattering probes, an important step toward making holographic instruments more mainstream, and is beneficial for cloud research. However, this comes at the cost of reduced capabilities to the previous design, especially a significant reduction in sample volume.

For the tethered balloon system, which moves slowly (say 1-1.5 m/s) this is acceptable, as it allows for a hologram every 20–30 cm (at 5Hz).  However, for an aircraft moving at much higher speeds(100-150 m/s), such a system would not be comparable to a scattering probe, nor to the existing holographic probes.

When comparing SmHOLIMO to HOLIMO, the mode of the cloud size distribution is below 10 microns—where HOLIMO tends to miss most of the droplets. This is a well-known limitation of previous holographic instruments and is more pronounced near the cloud base where the smallest droplets are. This effect should reduce as we go higher up and the mode shifts to larger sizes, say above 15 microns, where the differences might not be as drastic.

Given these considerations, I think the changes to the instrument design merit publication, but its scope is primarily limited to cloud droplet size distribution measurements on slow-moving or stationary platforms. While the authors mention that the instrument is specifically designed for a tethered balloon system, this point should be made more visible.

The paper is well written and easy to follow, but since this is an instrument design paper, the limitations of the instrument should be presented more clearly. Below are more specific comments:

1. Sizing uncertainty – With a pixel size of 1.85 microns, the sizing uncertainty is 1.36 microns (line 169), which introduces significant errors for smaller droplets near the resolution limit. For 3.7-micron droplets, this corresponds to a ~35% error, and for 5.55-micron droplets, it’s around ~25%. This should be accounted for, especially when analyzing the left-hand tail of the droplet size distribution. Additionally, considering a horizontal wind speed of 2 m/s (Figure 4d) and a pulse width of 220 ns, the particle displacement in a hologram would be ~0.4 microns—small but potentially contributing to the sizing uncertainty for the smallest droplets. 

2. Hologram image quality – The hologram quality is known to degrade due to speckle noise, particularly at higher droplet concentrations (Hui Meng et al., J. Opt. Soc. Am. A, 1993). This could become an issue when measuring clouds with high number concentrations, such as congestus or deep cumulus. It would be useful to add a sentence acknowledging this limitation and providing a rough estimate on limits for number concentrations.

3. Regarding Figure 5- Comparing SmHOLIMO to HOLIMO: It is unlikely that both instruments will measure all the droplets in their smallest size ranges. Therefore, in the 6–10 micron range, SmHOLIMO is expected to detect more droplets than HOLIMO. But It is surprising to see that concentration in this range for HOLIMO matches or in some cases is larger than that of SmHOLIMO. It looks like the bins are not the same for both instruments. Is this an artifact of the bin sizes? Showing both plots with identical bins would provide a clearer one-to-one comparison. Similarly, the particles in the smallest bins for SmHOLIMO are likely to be undercounted. A comparison with a forward scattering probe can be useful for the future, although forward scattering probes have their limitations in terms of sizing and particle counts as well.